Adaptive Control of Noncanonical Neural-Network Nonlinear Systems With Unknown Input Dead-Zone Characteristics

Most of the available results on adaptive control of uncertain nonlinear systems with input dead-zone characteristics are for canonical nonlinear systems whose relative degrees are explicit and for which a Lyapunov-based backstepping design is directly applicable. However, those results cannot be applied to noncanonical form nonlinear systems whose relative degrees are implicit and for which a Lyapunov-based backstepping design may not be applicable. This article solves the adaptive control problem of a class of noncanonical neural-network nonlinear systems with unknown input dead-zones. A complete solution framework is developed, using a new gradient-based design which is applicable to noncanonical nonlinear systems with input dead-zones. Signal boundedness of the closed-loop system and the desired tracking performance are ensured with the developed control schemes. Their effectiveness is illustrated by an application example of speed control of dc motors. This article can be readily extended to handle general parametrizable noncanonical nonlinear systems with unknown dynamics and input dead-zones, to solve such an open problem.